Sector level cost of equity in African financial markets

Emerging Markets Review 10 (2009) 257–278

Contents lists available at ScienceDirect

Emerging Markets Review j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / e m r

Sector level cost of equity in African financial markets Bruce Hearn a,b, Jenifer Piesse b,c,⁎ a b c

Sir John Cass Business School, United Kingdom King's College London, United Kingdom University of Stellenbosch, South Africa

a r t i c l e

i n f o

Article history: Received 15 July 2009 Received in revised form 10 September 2009 Accepted 11 September 2009 Available online 19 September 2009 JEL classification: G12 G15 O16

a b s t r a c t This paper assesses the effectiveness of Liu (2006) metrics in measuring illiquidity within a multifactor CAPM pricing model. Costs of equity are estimated using this model for the major sectors within Africa's larger equity markets: Morocco, Tunisia, Egypt, Kenya, Nigeria, Zambia, Botswana and South Africa. In all countries, the cost of equity is found to be highest in the financial sector and lowest in the blue chip stocks of Tunisia, Morocco, Namibia and South Africa. At an aggregate level, Nigeria and Zambia have the highest cost of capital. © 2009 Elsevier B.V. All rights reserved.

Keywords: Africa Capital asset pricing model Liquidity Emerging financial markets

1. Introduction The establishment and development of equity markets across African since the demise of the Cold War and subsequent restructuring of global capital flows has been driven by the need of many countries to attract foreign investment. Foreign Direct Investment (FDI) and portfolio investment are essential to supplement low domestic savings rates and are generally encouraged, despite political concerns about potential loss of sovereignty of national assets and vulnerabilities associated with financial contagion. African securities markets have achieved significant levels of institutional development during the last decade and strive to provide attractive and competitive venues for firms seeking to raise funds for much needed industrial and development projects. However, extreme illiquidity and segmentation are major

⁎ Corresponding author. Department of Management, King's College London, 150 Stamford St, London SE1 9HN, United Kingdom. Tel.: +44 207 848 4164. E-mail address: [email protected] (J. Piesse). 1566-0141/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.ememar.2009.09.002

258

B. Hearn, J. Piesse / Emerging Markets Review 10 (2009) 257–278

concerns for both potential investors and firms as they attempt to source cheap capital and diversify ownership through domestic listing (Lesmond, 2005; Hearn and Strange, 2009). Africa is a particularly interesting region in which to study securities markets given the current drive towards integration that is being actively pursued by regional bodies such as the New Africa Partnership for Development (NEPAD) and African Stock Exchanges Association (ASEA). There is a wide variety of markets at very different levels of development, from South Africa, which is the largest and most developed to the fledgling markets of Lusaka (Zambia) and Botswana. Contrasting levels of regulation and regulatory enforcement are reflective of the considerable variation in development amongst the continent's financial markets. Another factor that adds an intriguing dimension is the mix of legal regimes, with North and some West African markets influenced by French civil code (La Porta et al., 2008), as opposed to English common law in many other countries. Finally, a distinguishing feature in East Africa is the development underway in Kenya, which is preparing to act as a regional integrated hub market. By its nature, liquidity is a difficult concept to define, largely due to its ability to transcend a number of properties associated with market transactions, such as tightness, depth, resiliency (Lesmond, 2005) and information (O'Hara, 2003). Empirically defined constructs designed to capture liquidity centre on measuring direct trading costs, such as tightness, by the bid–ask spread (quoted or effective) and indirect trading costs that are linked to depth and resiliency, which are often represented by price impact measures. The lack of reliable and consistent bid–ask quotes in many emerging markets suggests the use of market activity proxies in capturing liquidity, and there is little consensus regarding the appropriateness of using common measures such as turnover and, more recently, the price impact variable developed by Amihud (2002) (Lesmond, 2005). However, the importance of including a measure of liquidity in pricing models and cost of equity analysis is supported by the poor performance of single-factor pricing models, particularly in emerging markets. Collins and Abrahamson (2006) provide costs of equity estimates for a variety of African markets but the analysis falters on the various forms of one-factor relationships used to model industry sector time series. The presence of severe illiquidity suggests a high degree of price rigidity, which lowers both variances and covariances between series (Hearn and Piesse, 2009), adding a significant bias in the betas, or their proxies, in CAPM type pricing models. Equally, costs of equity estimated through standard onefactor CAPM models (Correia and Uliana, 2004) fail to take into account the well documented effects of size and liquidity in explaining the cross section of returns (Martinez et al., 2005). Similar concerns relate to a study by Mishra and O'Brien (2005) that estimates the cost of equity using a two factor model at the firm level for 16 emerging markets that are included in the S&P Emerging Market Database (EMDB) (formerly IFC EMDB). This takes into account the market portfolio (MSCI World) and a political risk factor that relates the volatility of individual stocks to the volatility of the market. However, the implicit assumption that emerging markets are integrated with the global market portfolio and the omission of liquidity risk, which largely explains political risks, are particular sources of concern (Lesmond, 2005). While the literature on the importance of liquidity has developed over the past decade, research on liquidity risk and its applications is much more recent. Pastor and Stambaugh (2003) find evidence that leveraged investors that face solvency constraints require higher expected returns for holding assets that are difficult to sell when aggregate liquidity is low. Furthermore, stocks with a higher sensitivity to aggregate liquidity generate higher returns than low-sensitivity stocks, suggesting that liquidity is an important variable in asset pricing. Additionally a serious caveat in asset pricing and the ability to capture the cross section of stock returns arises from the inability of either the traditional one-factor CAPM or its three-factor counterpart, including sixe and book-to-market factors as proposed by Fama and French (1993) (Liu, 2006). Liu (2006) and Daniel and Titman (1997) find considerable evidence of the limited explanatory power of the Fama and French model to capture the cross section of asset returns. Martinez et al. (2005) also present evidence of the limited explanatory power of the Fama and French three-factor model, although there is some evidence of some explanatory power in retaining the size factor. Finally, Jun et al. (2005) find evidence that stock returns in emerging markets are positively related to liquidity. In addition to questions regarding the benefits of the book-to-market variable there are serious limitations to getting consistent accounting book values for firms in emerging markets. Emerging stock markets are highly skewed with many dominated by a handful of large firms while the rest are small and medium sized enterprises. Thus, a size factor should be retained within the pricing model to explain the cross section of returns. This study finds evidence that the liquidity and size factors are significant in

B. Hearn, J. Piesse / Emerging Markets Review 10 (2009) 257–278

259

explaining the cross section of returns and outperforms the traditional CAPM. Given that a liquidity factor better captures the risks attributable to firm distress and solvency issues than the additional book-tomarket valuation factor of the Fama and French (1993) model (Liu, 2006) this study augments the standard CAPM with both size and liquidity factors. The success of this multifactor CAPM provides support for the continued use of the risk–return paradigm in asset valuation. This paper is structured as follows. Section 2 has three parts: the first gives an overview of the institutional features of Africa's markets; the second introduces the liquidity measures and their construction; and the third discusses data specific issues. Section 3 outlines the three-factor size and liquidity augmented CAPM and Section 4 presents the empirical results. The final section concludes and outlines the impact these finding may have on development policy. 2. Equity markets and liquidity measurement 2.1. Securities markets The principal characteristics of these markets are summarised below and in Tables 1 and 2 (see Piesse and Hearn (2005) for an extended discussion of African stock markets). 2.1.1. North Africa The Egyptian stock exchange is one of the oldest and was formed by the integration of the Alexandria Stock Exchange, established in 1888, and the Cairo Stock Exchange, established in 1903. Trading is

Table 1 Market capitalisation and turnover profiles, 2008. Source: Compiled by authors from Bloomberg. North Africa

West Africa

East Africa

Southern Africa

Egypt

Tunisia

Morocco*

Nigeria

Kenya

South Africa†

Namibia**

Zambia

Botswana

302

53

78

234

46

373

8

24

20

Proportion market capitalisation to total (%) Top 1 7.43 12.51 Top 5 29.64 43.56 Top 10 43.58 65.23 Top 20 59.69 88.20

27.55 57.81 74.29 88.88

8.40 27.08 44.38 64.33

21.02 63.67 78.79 94.31

8.31 30.36 45.44 62.31

55.55 99.12 100.00 –

18.12 73.22 96.20 98.50

22.29 77.65 92.87 100.00

Proportion turnover value to total (%) Top 1 11.50 9.69 Top 5 36.81 38.19 Top 10 55.31 61.98 Top 20 78.67 86.51

19.42 58.92 78.00 92.01

9.00 36.45 52.66 69.59

20.44 56.73 74.76 95.78

12.74 42.19 59.67 76.28

46.77 100.00 100.00 –

32.85 92.49 98.71 99.00

18.52 68.53 94.55 100.00

Sector concentration by market capitalization Financials 24.72 57.38 Comm. 18.22 0.31 Basic materials (mining etc) 14.63 3.89 Consumer-cyclical 5.00 12.16 Consumer non-cyclical 6.59 8.92 Diversified 1.57 12.51 Energy 1.03 0.38 Industrial 18.86 4.45 Technology 0.12 – Utilities 0.18 –

42.04 27.55 3.37 2.55 4.35 7.39 1.46 9.93 0.14 1.23

59.74 1.37 0.53 2.81 17.79 1.57 10.41 5.35 0.02 –

46.90 0.96 1.63 4.73 26.95 0.08 4.41 14.33 – –

28.11 14.09 23.86 6.52 6.63 4.27 9.50 5.58 0.31 0.00

70.45 – – – 21.12 8.42 – – – –

18.25 18.12 – 1.22 37.93 – 3.43 15.48 – 5.53

85.33 – – 3.04 8.84 – 2.65 0.07 0.05 –

Listed firms

Notes: (1) *Refers to Central Market and Block Trading Market. (2) †Refers to both Johannesburg Stock Exchange main board and the Alternative Investment Market, ALTx. (3) **Refers only to local primary listed stocks. 68% of stocks listed on NSX have their primary listings elsewhere.

260

Table 2 Contrast of market regulations and commissions. No. brokers

Market clearance procedures

Capital gains tax

Trading hours

Trading arrangements

South Africa Namibia

Common law Common law

101 6

Fully G30 compliant including custodial facilities. DVP undertaken T + 3

Exempt

Kenya

Common law

18

Exempt

Egypt

Civil code

146

Partial G30 compliant. DVP undertaken T + 3. Fully G30 compliant including custodial facilities. DVP undertaken T + 3.

8:25 am–9:00 am: pre-opening electronic call auction. 9:00 am–4:00 pm: continuous trading. 4:00 pm–6:00 pm: run-off 10:00 am–12:00 noon

Morocco

Civil code

15

Exempt

Tunisia

Civil code

24

Exempt

JSE SETS electronic trading system (SETS trading system has been in place at the London Stock Exchange and replaced the former JET system in 2002) Automated trading system — previously open outcry commenced by sounding a bell Electronic order matching system for Cairo and Alexandria Stock Exchanges (CASE) — the CASE trading system, or CTS Electronic continuous auction system. Secondary market trading is segmented into blocks: continuous market; block market (for large institutional trades); development market (for SMEs) Electronic order matching system

Nigeria

Common law

219

Zambia

Common law

3

Botswana

Common law

4

Custodial facilities provided by brokers with sufficient capitalisation. Mostly G30 compliant and DVP undertaken at T + 3 Partial G30 compliant. DVP undertaken T + 3.

Exempt

Exempt

Exempt Exempt

Listed securities market (on the Exchange): 11:30 am–15:30 pm

9:00 am–10:00 am: pre-opening// 10:00 am–11:30 am trading session 11:00 am–13:00 pm

9:00 am to 10:00 am: pre-opening// 10:00 am–13:00 pm trading session 10:00 am–12:00 pm

Call Over trading system was replaced in April 1999 by automated trading system (ATS) which serves as an electronic order matching system Automated trading system (continuous auction) Electronic call auction conducted in trading room within exchange building.

Note: (1) South Africa and Namibia adhere to Roman–Dutch civil code but commercial and securities regulatory law follows English common law.

B. Hearn, J. Piesse / Emerging Markets Review 10 (2009) 257–278

Commercial law

B. Hearn, J. Piesse / Emerging Markets Review 10 (2009) 257–278

261

electronic and between floor-based members of local brokerage firms and takes place in three markets: an OTC market, a Primary Dealers Bonds market, and a Listed Securities market. The latter has a pre-opening session from 9:45 am to 10:15 am followed by continuous daily trading between 10:30 am and 14:30 pm. A central depository exists to assist settlement, which is largely in compliance with G30 recommendations1 and a number of large, well capitalized, custodian banks support overseas investors (CASE website, 2009). Significant improvements have been made to market regulations and the universal adoption of corporate governance and accounting standards, which through their costly implementation caused a dramatic reduction in listings in 2004 (CASE website, 2009). However many of the listed firms are smaller in size and dominated by either family stakeholders or blockholders and only a small handful of firms such as Orascom have been able to meet the more stringent regulatory requirements and dual list shares or depository receipts on major overseas exchanges. The code of corporate governance established in 2003 was set up by a committee formed from CASE and the largest 10 companies to enshrine the best principles of OECD guidelines and despite the considerable progress that has been made there remains significant ambiguity in firm's interpretation of the various directives designed to ensure market transparency (Fawzy 2003). The Bourse de Casablanca, Morocco, was established in 1929. Trading is electronic with terminals located in the local brokerage community. Settlement is G30 compliant by MAROCLEAR, the national CSD established in 1998 (Bourse de Casablanca website, 2009). Trading is reported electronically to local market participants and to international data vendors such as Bloomberg and Reuters. This gives the market the opportunity to attract overseas investors. Stock market awareness is high and the exchange is used as a successful route for domestic flotation, although it also attracts significant retail and institutional investors. Corporate governance legislation in Morocco, alongside the rest of North Africa, has undergone a considerable modernization in line with the changes in political, economic and administrative governance institutions. However, formal legislation, in the form of a Moroccan Code of Corporate Governance Practices, has only very recently been enacted in February 2007 through the establishment of a National Commission of Corporate Governance in Casablanca (National Commission on Corporate Governance 2008). This has overseen the legal codification of corporate governance in the Kingdom, which largely follows the OECD best practice guidelines. The Bourse de Tunis was established in 1969. Trading is electronic and was introduced in 1996 with the assistance of Euronext Paris. The trading system is split into fixing and continuous systems, with the former handling small and illiquid securities and comprised of a series of sequential electronic call auctions (Bourse de Tunis website, 2009). Trading hours in the continuous market are 9:00 am to 14:10 pm in the months outside July, August and during Ramadan where hours are 8:30 am to 12:10 pm. Settlement is fully G30 compliant. Overall market regulation is well designed through considerable French and EU assistance and corporate governance, in line with Morocco, adheres to OECD best practice guidelines. However, there is little of a domestic stock market culture and only around 5% of finance raised by firms in 2007 was done so through the stock exchange (Zribi, 2008). 2.1.2. West Africa The Nigerian stock exchange is the largest and most active stock exchange in the West African region2 and was originally established in 1960 in Lagos with a mere 19 listed firms. The exchange now has several branches around the country including Kaduna (established in 1978) and Port Harcourt (established in 1980) (Nigerian stock exchange website, 2009). However serious concerns over the small and fragile formal business sector in Nigeria together with intense political lobbying caused the early closure of the 1 G30 relates to the Group of Thirty which is the most influential body to encourage the standardisation and improvement in global securities administration. Following a symposium in London in March 1989, the following recommendations were agreed: i) Brokers should match trades on day after deal date (T + 1); ii) Trade confirmation on trade day plus 2 days (T + 2); iii) Central Depository for safe keeping of shares; iv) Net basis settlement of cash and stock; v) Settlement takes place as delivery vs. payment or receipt vs. payment; vi) Settlement in same day funds; vii) Settlement effected on trade date plus 3 days (T + 3) 8; viii) Securities lending should be permitted; and ix) International securities numbering system must be adopted (ISIN code). 2 There are also much smaller stock exchanges in Ghana, Cote d'Ivoire (acting as a regional exchange for Francophone member states of Union Monétaire et Économique de l'Afrique de l'Ouest (UMEAO)), and Cape Verde Islands, but these are omitted from this study due to their size and extreme illiquidity (see Lavelle (2001) and Hearn and Piesse (2009) for a detailed discussion of these markets).

262

B. Hearn, J. Piesse / Emerging Markets Review 10 (2009) 257–278

country's second stock exchange in Abuja in 2001 with the institutions being transferred to support the newly created Abuja Commodities Exchange (Abuja Commodities Exchange website, 2009). Trading activity is concentrated in Lagos and is undertaken by an automated trading system (ATS) on a daily basis from 11:00 am to 14:00 pm. Settlement is partially G30 compliant, with a CSD created in 1992 and international custodian banks (Nigerian stock exchange website, 2009). The network of 219 brokers to support the 234 listed firms ensures it is the largest market in West Africa and although trading activity and capitalization is less concentrated than neighbouring markets the financial sector accounts for 59.74% of market capitalization, as seen in Table 1. However, although internationally recognised auditors and accountants are present there are considerable differences in the use of International Accounting Standards (IAS) in listed firms. Many poorly capitalized firms are unable to afford the high fixed costs of professional auditing and there are many ambiguities concerning the use and overlap of Nigerian accounting standards and IAS. Consequently, corporate governance is a costly luxury for those few firms that have sufficient capitalization to comply. 2.1.3. East Africa The Nairobi Stock Exchange (NSE) was first established as an informal share market during the 1920s before being incorporated in 1954. Since 1994, the market has occupied the same premises as the CSD and The Nation, which is the business journal that disseminates trading information. Trading takes place daily between 10:00 am and 12:00 by a central electronic book entry system, and is limited to the floor of the exchange. The market is dominated by blockholders and smaller retail investors and free float percentages are low3. Order flow to the market is by a small network of licensed stock brokers and their regional affiliates. Investors are required to establish both a trading account with the broker and a separate individual account at the central depository. Public releases of shares in the primary market and IPOs are managed through local investment banks, with the Capital Markets Authority (CMA) responsible for regulation and supervision. The CMA acts to enforce market regulation and corporate governance is modelled loosely on prevailing OECD standards although there remains considerable ambiguity in its interpretation (Barako et al., 2006). 2.1.4. Southern Africa The Johannesburg Securities Exchange in South Africa is the oldest and largest market and was established in 1887. A sophisticated electronic trading system has been adopted since the end of the former open outcry system in 1996. This system was extended to become a regional trading system linking the integrated market of Namibia in 1998 and upgraded in 2002 under the guidance of the London stock exchange to the current Shares Electronically Traded System (SETS). There is a central depository, the Southern African Financial Instruments Clearing and Settlement System (SAFICAS), which is based on technology used in the Swiss stock exchange. There are high levels of corporate governance, following the King I and II reports4, and international regulatory standards (JSE website, 2009). Namibia is similar in market organization, having shared a colonial past and the associated common legal heritage and institutions. Both South Africa and Namibia are members of the Common Monetary Area (CMA)5 and South African Customs Union (SACU) (Hearn and Piesse, 2002). The Botswana stock exchange was established in 1989 as an over-the-counter share market before trading was formalised in 1995 and now functions as a call auction daily from 10:00 am to 12:00 pm. The exchange has 20 listed companies, with the financial sector accounting for over 85% of market

3

The proportion of issued shares available to the public and not held by incumbent block holders. The King Reports that regulate corporate governance practices in South Africa is very similar to the UK Cadbury Report and the US Sarbanes–Oxley Act (South African Institute of Directors, 2009). 5 Common Monetary Area (CMA) countries include Namibia, Swaziland and Lesotho as well as South Africa. Member states currencies are pegged to the Rand and form an economic union. 4

B. Hearn, J. Piesse / Emerging Markets Review 10 (2009) 257–278

263

capitalization. Order flow is through four local brokers and the market provides limited risk diversification opportunities for the small institutional investment community, mostly made up of pension funds (Jefferis, 1995). A fully G30 compliant CSD facilitates payment and settlement. Due to geographical proximity, the Botswana market is heavily influenced by the corporate governance standards in South Africa. This is further reinforced by a number of locally listed blue chip South African firms that follow a high quality regime of governance, accounting and auditing standards. The Zambian stock exchange was established in Lusaka in 1994 with technical assistance from the International Finance Corporation (IFC) and the World Bank. Trading is by delocalised electronic system, with three brokers. There is a pre-opening call auction between 9:00 am and 10:00 am followed by a continuous daily auction from 10:00 am to 13:00 pm. A fully G30 compliant CSD facilitates payment and settlement and the Securities and Exchange Commission (SEC) has regulatory oversight of the market. However, the evidence in Table 1 suggests that the market is highly concentrated with one stock alone accounting for 32.85% of capitalization and 18.12% turnover. Corporate governance in Zambia is still in its infancy with indigenous firms making up the majority of local listings and the few foreign firms, mostly from South African, list a very small proportion of their overall shares (Old Mutual, 2009). In this market, the formal sector is small and concentrated on the consumer goods, financial and communication industries, as shown in Table 1. In Zambia, the informal sector dominates the economy. 2.2. Liquidity constructs 2.2.1. The bid–ask spread and commission costs The data on the end of month bid and ask quotes were from Datastream for Morocco, Egypt and South Africa and from Bloomberg for Tunisia. Data were unavailable for Namibia, Botswana, Nigeria and Zambia. There is considerable variation in the length of intraday data with Morocco and Tunisia available for over 15 years and Egypt and South Africa from 2000. Because of inconsistencies between the various data sources some was obtained directly from the markets. The bid–ask spread is calculated using the average of the available monthly quotes and incorporates at a minimum a single month's quote for that month. The average bid–ask spread spanning the quarter is used for the estimate of the spread. This minimizes outliers and averages out the highs or lows resulting from monthly sampling. Following Lesmond (2005) bid–ask spreads that exceed 80% are trimmed as these are potentially errors. The monthly quoted spread is defined as: 

 ðAskM −BidM Þ ðAsk M−1 −BidM−1 Þ + Quoted spreadM = 1 = 2 ðAskM + BidM Þ = 2 ðAsk M−1 + BidM−1 Þ = 2

ð1Þ

2.2.2. Liu (2006) measure Daily price and volume data are collected from Datastream. The measure is from Liu (2006) and defined as LMx which is the standardised turnover-adjusted number of zero daily trading volumes over the prior x months (x = 1,6,12), that is  1 = x month turnover LMx = ðNumber of zero daily volumes in prior × monthsÞ + Deflator

 21x  NoTD

ð2Þ where x month turnover is the turnover over the prior x months. This is calculated as the sum of the daily turnover over the prior x months, which is the ratio of the number of shares traded over the number of shares outstanding at the end of the day. NoTD is the total number of trading days over the prior x months and Deflator is chosen such that, 1

0〈

. ðx month turnoverÞ

Deflator

〈1

ð3Þ

264

B. Hearn, J. Piesse / Emerging Markets Review 10 (2009) 257–278

for all stocks6. Given the turnover adjustment (the second term in the brackets in Eq. (2), two stocks with the same number of zero daily trading volumes can be distinguished: the one with the larger turnover is more liquid. Thus the turnover adjustment acts as a tie-breaker when sorting stocks based on the number of zero daily trading volumes over the prior x months. Because the number of trading days per month can vary from 15 to 23, multiplication by the factor (21 × /NoTD) standardises the number of monthly trading days to 21, which makes the liquidity measure comparable over time. The Liu measure, designated LM1 with 1 reflecting the period of measurement i.e. one month, can be interpreted as the turnover-adjusted number of zero daily trading volumes over the prior 21 trading days, which is the approximate average number of monthly trading days. The liquidity measure, LMx is calculated at the end of each month for each individual stock based on daily data. 2.3. Data: sources Daily stock closing, bid and ask prices, total number of shares outstanding, traded volumes, dividend per share in local currency and converted into sterling were obtained for Egypt, Morocco and South Africa from Datastream. These variables are from Bloomberg and the national stock exchanges for Tunisia, Nigeria, Botswana, Zambia and Namibia. Using these data, the daily return variance (volatility), market capitalisation, defined as the total number of shares outstanding multiplied by daily closing price, and various liquidity constructs were calculated. The total returns series for each stock were from Datastream for Kenya, Morocco, Egypt and South Africa. Those for Nigeria, Tunisia, Zambia, Botswana and Namibia were constructed using the Standard & Poors method, assuming dividend reinvestment and allowing for dividends, stock splits, rights issues and other corporate activity that may affect a stocks intrinsic value. Exchange rate and UK-Gilt/Treasury yield data, which represents the risk free rate adjusted to take account of monthly excess returns rather than the quoted equivalent annualised rates, are from Datastream. The conversion of the total returns series and prices into sterling and the use of UK-Gilt/Treasury yield rate assume long term parity between individual domestic currencies and sterling. In many cases companies were deleted from the sample owing to either data inconsistencies or the lack of certain variables that make total return calculations impossible. Nigeria is one example where there are 234 listings yet 60 of these do not have data and a further 45 firms have incomplete data and consequently the sample size for Nigeria is 129 firms. 2.4. Data: summary statistics of liquidity measures The skewness and considerable differences in liquidity in these equity markets are shown in Table 3. This provides a contrast of the mean cross-section values for daily percentage zero returns, stock prices, traded volumes, market capitalisation and bid–ask spreads for the firms within the overall market, the top tier stocks, as ranked by market capitalisation, and for the largest sectors defined in Table 1. The top tier stocks for Egypt, Morocco, Nigeria and South Africa contain 10 firms, while those for Kenya and Tunisia have 5 firms and Botswana and Zambia 3 firms. There is clear evidence of a size and liquidity effect in all the markets, with the mean cross-section capitalizations of the top tier stocks in each case frequently being larger than the aggregate market by several orders of magnitude. Similarly, there are considerable differences between the top tier stocks and the aggregate market in terms of size of the bid–ask spread. For example, the mean bid–ask spread of the top tier is just 3.40% of the mean for the overall market for South Africa. These differences within the markets are further reflected by the differences between industrial sectors. Even taking into account that only the largest sectors, as ranked by market capitalisation in Table 1, have been used there are considerable contrasts. Interestingly, in all markets although the financial sector forms the highest profile listings and a steady source of blue chip listings to the exchanges their liquidity profile is poor as compared to the top tier stocks and often only marginally better than the aggregate market. Percentage daily zero returns and, where available, bid–ask spreads, reveal that in all cases liquidity is only marginally less than the overall market and considerably lower than for the top tier stocks. However, the greatest degree of illiquidity that is consistent across all groups of firms is in Botswana, 6

Following Liu (2006) a deflator of 1000 is used in constructing estimates for LM1.

B. Hearn, J. Piesse / Emerging Markets Review 10 (2009) 257–278

265

Table 3 Summary statistics. Source: Compiled by authors from Bloomberg, Datastream and National stock exchanges. Local market Country

Start

No. firms by trading activity

North Africa Egypt 1998 Financial Basic material Consumer non-cyclical Industrial Top 10 121 Morocco

1993 Financial Consumercyclical Diversified Industrial Top 10 40

Tunisia

1991 Financial Consumercyclical Consumer non-cyclical Top 5 37

West Africa Nigeria 2002 Financial Consumer non-cyclical Energy Top 10 129

East Africa Kenya 1995 Financial Consumer non-cyclical Industrial

£UK equivalent

Zero return (%)

Price

Volume (thousands)

Market capitalization (millions)

Price

Market capitalization (millions)

Bid–ask spread (%)

48.09 [50.55] 46.96 [52.17] 54.64 [57.14] 46.07 [45.94] 21.08 [21.21] 50.92 [53.24] 65.00 [65.73] 72.45 [82.61] 52.10 [48.33] 68.42 [66.66] 43.82 [39.55] 66.43 [66.67] 59.45 [60.35] 60.98 [59.09] 72.33 [69.69] 50.82 [49.09] 64.97 [66.04]

26.22 [23.58] 56.37 [21.45] 93.74 [123.34] 26.26 [18.46] 91.39 [33.45] 42.05 [44.83] 449.98 [393.11] 106.88 [83.25] 783.13 [733.94] 594.01 [525.20] 663.27 [636.16] 616.07 [635.06] 25.82 [22.63] 16.44 [13.55] 43.23 [41.16] 25.40 [21.65] 34.65 [32.02]

8044.67 [4534.04] 3345.33 [2711.79] 399.76 [234.89] 2034.54 [1159.02] 13,247.49 [12,518.37] 6027.75 [4760.21] 2795.61 [1361.85] 2185.28 [434.65] 287.65 [108.00] 57.69 [17.10] 4568.47 [1507.30] 7071.77 [5205.65] 729.80 [593.82] 224.79 [93.57] 59.08 [27.07] 324.47 [222.34] 1084.96 [835.74]

1,026.78 [322.50] 239.66 [101.02] 88.48 [63.93] 382.92 [180.55] 832.02 [357.63] 1,535.97 [789.51] 3,613.09 [2,444.36] 759.67 [473.73] 8,906.95 [9,134.44] 5,647.75 [5,618.73] 12,228.62 [8,519.74] 4,045.14 [3,077.26] 130.50 [107.61] 46.57 [40.30] 104.78 [76.46] 254.76 [264.31] 58.59 [48.28]

2.75 [2.74] 5.52 [2.75] 11.11 [10.74] 2.51 [1.85] 9.07 [4.74] 4.31 [4.86] 28.87 [24.67] 6.84 [5.15] 50.38 [45.67] 38.25 [32.88] 42.63 [40.36] 39.62 [39.75] 14.29 [11.67] 8.63 [6.97] 24.56 [21.68] 46.26 [41.48] 18.93 [16.02]

99.77 [38.10] 23.59 [11.43] 9.51 [8.85] 37.78 [19.91] 82.33 [51.09] 162.45 [117.44] 241.15 [163.96] 49.03 [29.58] 569.83 [557.31] 360.83 [349.10] 785.21 [528.93] 263.65 [200.44] 72.44 [57.28] 32.05 [21.41] 50.89 [48.39] 131.49 [122.81] 103.70 [95.10]

4.14 [2.90] 2.59 [2.26] 13.95 [14.76] 4.53 [4.89] 0.77 [0.74] 5.62 [6.06] 0.83 [0.08] 0.79 [0.04] 1.03 [0.00] 0.80 [0.00] 0.17 [0.00] 0.54 [0.00] 2.08 [1.99] 3.70 [2.42] 2.85 [1.99] 6.40 [6.43] 53.37 [69.13]

53.61 [58.55] 57.94 [60.17] 41.35 [39.67] 41.96 [43.18] 63.67 [66.94]

32.45 [39.13] 17.03 [16.41] 82.58 [79.76] 12.04 [7.24] 18.76 [18.44]

2,032,900.25 [1,054,456.35] 182,111.90 [122,372.26] 36,095.08 [24,145.01] 1,537,138.03 [920,356.09] 2,631,207.67 [1,373,235.44]

53,641.83 [27,970.54] 1,956.04 [2,112.87] 32,299.59 [34,046.92] 123,079.19 [51,593.87] 30,129.70 [25,008.37]

0.13 [0.16] 0.07 [0.06] 0.36 [0.32] 0.05 [0.02] 0.08 [0.07]

300.68 [141.32] 8.36 [8.64] 138.49 [137.45] 540.25 [214.07] 132.49 [102.05]

–

47.84 [49.65] 69.28 [70.56] 70.55 [72.85]

29.67 [17.99] 74.88 [76.21] 33.41 [30.25]

116,329.73 [35,845.90] 32,676.73 [9836.90] 6017.40 [1590.00]

5,249.66 [19,340.63] 3,368.45 [12,164.60] 5,446.07 [25,108.37]

0.24 [0.19] 0.65 [0.60] 0.28 [0.24]

41.05 [148.60] 29.18 [151.82] 39.46 [227.83]

4.08 [4.03] 2.96 [2.83] 3.07 [2.70]

– – – –

(continued on next page)

B. Hearn, J. Piesse / Emerging Markets Review 10 (2009) 257–278

266 Table 3 (continued)

Local market Country

Start

£UK equivalent

No. firms by trading activity

Zero return (%)

Price

Volume (thousands)

Market capitalization (millions)

Price

Market capitalization (millions)

Bid–ask spread (%)

Top 5

39.76 [40.00] 60.31 [61.63]

39.08 [14.86] 45.01 [41.10]

91,439.09 [28,159.00] 174,106.53 [59,870.70]

– 3,582.61 [16,327.59]

0.38 [0.18] 0.38 [0.36]

154.17 [446.72] 31.88 [145.32]

2.92 [3.24] 3.50 [3.62]

1717.81 [1627.15] 1,801.78 [1,315.63] 4,367.54 [3,671.98] 1,628.92 [1,300.37] 985.29 [796.64] 4,130.46 [2,636.55] 1,180.60 [907.97] 10,875.53 [8,845.80] 1,832.36 [1,375.33] 13.86 [1.58] 8.28 [1.65] 224.09 [150.65] 630.00 [671.11] 335.13 [211.32] 291.48 [216.36] 2,182.70 [1,777.02] 926.08 [238.55] 877.18 [330.62] 116.50 [39.66] 981.18 [636.55]

615,956.42 [622,400.30] 103,311.43 [78,310.76] 268,319.65 [200,649.96] 279,201.93 [174,665.35] 246,512.39 [247,604.33] 42,129.91 [41,485.32] 168,019.15 [143,755.51] 436,124.70 [442,050.81] 1,846,339.16 [1,635,892.97] 725.99 [189.1] 1658.26 [746.24] 9427.84 [6400.21] 1293.33 [569.17] 5297.94 [2634.31] 11,106.26 [7466.05] 27,515.11 [427.6] 4439.44 [1244.82] 1892.49 [78.87] 6231.29 [1159.1] 37,883.91 [4123.63]

714,336.90 [5,467,097.31] 1,240,306.64 [7,513,715.89] 1,706,137.77 [13,536,731.66] 466,612.01 [3,142,816.10] 255,289.09 [1,827,363.15] 3,121,798.49 [20,758,811.13] 305,975.59 [2,217,807.39] 7,159,093.22 [50,766,503.99] 698,166.17 [4,991,884.70] 207.70 [156.5] 206.66 [147.90] 50,208.04 [31,531.52] 51,656.72 [41,913.88] 14,133.67 [11,133.75] 37,295.78 [23,959.62] 215,983.31 [101,942.25] 245,914.59 [62,472.12] 119,399.79 [22,125.33] 415,446.40 [119,251.55] 157,404.07 [48,538.56]

1.52 [1.54] 1.55 [1.52] 3.46 [3.01] 1.39 [1.28] 0.86 [0.79] 3.31 [2.26] 1.00 [0.86] 8.66 [6.82] 1.56 [1.41] 0.87 [0.17] 0.59 [0.17] 21.93 [16.59] 67.11 [73.37] 32.04 [22.37] 29.55 [23.29] 0.38 [0.25] 0.13 [0.03] 0.12 [0.06] 0.02 [0.01] 0.16 [0.14]

548.30 [4,672.59] 931.35 [6,494.58] 1,281.98 [10,420.37] 356.69 [2,662.86] 193.51 [1,539.38] 2,340.72 [17,555.67] 232.80 [1,921.39] 5,420.18 [43,196.07] 529.35 [4,215.62] 19.71 [15.99] 18.35 [14.55] 4664.87 [3331.14] 5,345.24 [4,316.3] 151,475.21 [97,449.5] 3,530.67 [2,532.89] 32.39 [17.63] 34.17 [7.87] 16.60 [3.95] 58.70 [15.60] 22.53 [6.61]

–

East Africa

37

Southern Africa South 2000 Financial 48.42 Africa [47.93] Communication 41.17 [39.92] Basic material 39.49 [41.88] Consumer44.69 cyclical [44.28] Consumer 45.45 non-cyclical [45.45] Energy 43.88 [44.44] Industrial 46.81 [45.65] Top 10 9.35 [8.09] 273 45.44 [45.11] Namibia* 1995 Financial 92.96 [93.8] 7 92.53 [93.48] Botswana 1996 Financial 89.43 [90.00] Consumer 89.40 non-cyclical [90.48] Top 3 86.13 [87.30] 19 90.49 [91.15] Zambia 1997 Financial 91.74 [92.86] Consumer 86.02 non-cyclical [87.27] Industrial 83.87 [84.42] Top 3 79.44 [80.95] 18 89.75 [90.91]

– – – – – – 0.40 [0.60] 10.07 [10.38] – – – – – – – – – – –

Note: (1) *Indicates Namibian domestic market of 7 locally listed firms. Remaining 22 Namibian firms have primary listings in South Africa and are considered South African.

Namibia and Zambia. Zambia has a value of 89.75% for the aggregate market, falling to 79.44% for the top 3 stocks and Botswana falls from 90.49% for the aggregate market to 86.13% for the top 3 stocks. Namibia is the most illiquid markets with the percentage daily zero returns over 92% for the aggregate local market indicting severe price rigidity. These results form the basis of the critique in this paper regarding the estimation of the cost of equity in African markets using standard techniques that do not take account of either issues relating to market segmentation of the severe illiquidity that exists. The severity of the

B. Hearn, J. Piesse / Emerging Markets Review 10 (2009) 257–278

267

illiquidity reduces the variances and covariances of the stocks and leads to the conventional application of CAPM market betas inaccurately in these emerging markets. 3. Empirical model: size and liquidity augmented CAPM In the spirit of the three-factor CAPM model of Fama and French (1993) this work follows the recent work of Martinez et al. (2005) and Shum and Tang (2005) by modifying the augmented factors to take account of size and liquidity effects that offer improved performance in capturing anomalies across the cross section of stock returns especially prevalent in emerging markets. Thus, in addition to market excess returns, the model is augmented by the excess returns attributed to size (SMB) and to illiquidity (ILLIQ). The market, size and liquidity factors used in the CAPM are formed from the universe of available stocks and sorted into equally weighted portfolios with rebalancing undertaken in December each year from 2002 to 2008. All stocks are assumed to be held continuously for a further year following rebalancing. The market portfolio itself is the simple arithmetic mean of the cross section of total returns in the universe. The universe is sorted each December, first by each stock's market capitalization into three size ranked portfolios, “Small”, “Medium”, and “Big”, and then each of these is further sorted into another three portfolios based on the liquidity measure. The size factor is formed from the cross-section mean returns of the small-size portfolio minus the big size portfolio and is referred to as the SMB (small minus big) factor, following Fama and French (1993). Given the Amihud liquidity measure, which focuses on the impact on a stock's price of trading activity, outperforms the turnover construct in capturing the effects of liquidity within the sample markets this is used in the ranking of s stocks into portfolios based on their relative liquidity. This liquidity factor is based on the mean of each of the three “High” illiquidity portfolios within each of the size portfolios, minus the mean of the “Low” illiquidity portfolios and is referred to as the HML (high minus low), following Liu (2006). The market variable is problematic due to the lack of appropriate regional benchmarks in sub-Saharan Africa (SSA) and it is further complicated by the unreasonable assumption that full market integration would impose on these highly segmented markets. Thus, the North African universe of markets includes Egypt, Tunisia and Morocco. The South African universe includes South Africa and Namibia as they are both members of a currency union and also share a trading platform and legal and regulatory systems. Finally, a sub-Saharan universe includes Nigeria, Kenya, Zambia and Botswana. Market universes formed in this manner minimise the difficulties of including extremely heterogeneous markets within the common integrated market assumption. Once the three factors have been constructed the three-factor CAPM can be restated as the expected return on a risky portfolio p, in excess of the risk free rate E(Rp) – Rf is a function of (i) excess return on the market portfolio, Rm – Rf ; (ii) the difference between the return on a portfolio of small-size stocks and of large-size stocks, SMB; and (iii) the difference between the return on a portfolio of high illiquidity stocks and of low illiquidity stocks, ILLIQ. Therefore, the expected excess returns on a portfolio p of emerging market stocks can be written as Eðrpt Þ−rft = βp ½Eðrmt Þ−rft  + si EðSMBÞ + hi ðHMLÞ

ð4Þ

The equilibrium relation of the Fama and French (1993) three-factor model is stated in terms of expected returns. In order to test the model with historical data, it is necessary to transform Eq. (4) to the following estimating equation: rit −rft = αi + βi ðrmt −rft Þ + si SMBt + hi HMLt + εit

ð5Þ

where the variables are described above and εp,t is an iid disturbance term. 4. Results 4.1. Summary statistics of the size–liquidity sorted portfolios The dispersion of stocks on a market by market basis between the nine size–illiquidity sorted portfolios is shown in Table 4. These have been generated for the markets of North and South Africa. An additional

268

B. Hearn, J. Piesse / Emerging Markets Review 10 (2009) 257–278

Table 4 Average number of stocks in each of the 9 size–illiquidity portfolios sorted by nationality by year in period: 2002–2008. Portfolio

S/L

S/M

S/H

M/L

M/M

M/H

B/L

B/M

B/H

6.00 2.00 4.00 12.00

9.00 3.14 1.00 13.14

14.00 0.00 1.00 15.00

8.43 2.98 3.00 14.40

3.00 6.86 6.00 15.86

7.00 3.00 5.00 15.00

2.00 5.92 6.00 13.92

3.00 6.86 4.00 13.86

Africa (excl. RSA) 0.00 0.00 10.00 7.00 0.00 0.00 0.00 3.73 10.00 10.73

0.00 7.92 0.00 0.86 10.78

4.72 8.53 0.00 0.00 13.26

0.79 6.00 0.00 4.00 10.79

0.00 4.89 0.00 2.77 10.52

10.82 0.93 0.00 0.00 11.76

5.79 0.99 5.92 0.00 12.70

0.00 0.00 8.17 0.00 8.17

9.19 4.74 13.93

17.68 0.00 17.68

14.07 0.00 14.07

14.27 0.00 14.27

16.86 0.00 16.86

13.00 0.00 13.00

14.71 0.00 14.71

Market: North Africa Egypt 10.29 Morocco 0.93 Tunisia 2.00 Overall mean: 13.21 Market: sub-Saharan Nigeria Kenya Zambia Botswana Overall mean:

Market: South Africa South Africa 12.70 Namibia 0.00 Overall mean: 12.70

14.27 0.00 14.27

sub-Saharan Africa market variable has been generated based on the few prevailing regional benchmarks that exist, which excludes South Africa and Namibia, but includes Botswana, Zambia, Kenya and Nigeria. There is an even dispersion of stocks across all size–illiquidity sorted portfolios in South Africa, although the small number of Namibian stocks are all located in the small-size, high illiquidity portfolio. In the North Africa size–illiquidity portfolios there is a relatively even dispersion although Egypt and Tunisia tend to dominate the larger size portfolios while Morocco and Tunisia tend to dominate the more illiquid portfolio. The greatest dispersion occurs in the sub-Saharan Africa case. Nigeria and Zambia dominate the large-size portfolios while Kenya and Botswana are in small to medium size portfolios. Notably, Zambia and Botswana stocks are concentrated in the high illiquidity portfolios emphasising the severe illiquidity present. Descriptive statistics for all nine size–illiquidity portfolios, the mean industry portfolios, and the zerocost SMB and ILLIQ portfolios, formed under the assumption of no intrinsic arbitrage opportunities across component stocks, are in Tables 5,6 and 7. Table 5 shows that the average mean returns increase Table 5 Summary statistics for equally weighted monthly excess returns on 9 portfolios formed on size and illiquidity for period 2002 to 2008. Portfolio

S/L

S/M

S/H

M/L

M/H

B/L

0.0192 0.0520 0.4306 4.1632

0.0116 0.0378 − 0.2237 3.4270

0.0205 0.0546 −0.4099 4.5507

0.0258 0.0522 0.3817 2.8852

0.0095 0.0355 −0.6351 4.9243

0.0292 0.0906 0.1375 3.5922

0.0554 0.1645 5.1734 37.7099

0.0248 0.0507 0.4022 5.3684

0.0259 0.0711 −0.0279 4.4232

0.0232 0.0756 1.7870 13.2611

0.0548 0.3084 8.0741 70.4814

0.0207 0.0823 − 0.4230 3.4607

0.0214 0.0811 − 0.4548 3.3583

0.0220 0.0749 −0.6059 3.4804

0.0181 0.0804 − 0.2822 2.8291

0.0176 0.0785 − 0.6786 3.5500

0.0197 0.0755 − 0.7359 4.0560

Market: North Africa Mean Std. dev. Skewness Excess kurtosis

0.0200 0.0752 0.3693 3.5889

0.0171 0.0520 0.0929 2.6394

0.0012 0.0355 0.6626 4.3447

0.0227 0.0911 0.8201 4.9198

Market: sub-Saharan Mean Std. dev. Skewness Excess kurtosis

Africa (excl. RSA) 0.0414 0.0508 0.1098 0.1294 1.0623 4.2679 4.1811 28.4699

0.0078 0.0566 1.8479 7.9806

0.0345 0.0812 0.3547 3.3875

Market: South Africa Mean 0.0322 Std. dev. 0.0909 Skewness − 0.4408 Excess kurtosis 3.1252

0.0378 0.0835 − 0.4169 2.9430

Note: (1) Values in parentheses are standard deviations. (2) *Indicates ranking by market capitalization.

M/M

B/M

B/H

B. Hearn, J. Piesse / Emerging Markets Review 10 (2009) 257–278

269

Table 6 Summary statistics for valuation factors. SMB Market: North Africa Panel A: summary statistics for valuation factors Mean − 0.0211 Standard deviation 0.1098 Skewness 0.4079 Excess kurtosis 3.7403 Panel B: correlations for valuation factors SMB 100.00% ILLIQ − 1.16% MARKET 0.32% Market: sub-Saharan Africa (ex. RSA) Panel A: summary statistics for valuation factors Mean − 0.0076 Standard deviation 0.3723 Skewness − 5.5310 Excess kurtosis 44.9210 Panel B: correlations for valuation factors SMB 100.00% ILLIQ − 74.90% MARKET − 22.88% Market: South Africa Panel A: summary statistics for valuation factors Mean 0.0455 Standard deviation 0.1213 Skewness 0.1365 Excess kurtosis 3.2288 Panel B: correlations for valuation factors SMB 100.00% ILLIQ 53.04% MARKET − 1.64%

ILLIQ

MARKET

− 0.0445 0.1698 − 0.2049 4.3973

0.0167 0.0389 − 0.1117 2.8843

– 100.00% − 74.21%

– – 100.00%

− 0.0127 0.3911 5.0305 39.8079

0.0330 0.0526 0.7647 4.1513

– 100.00% 3.70%

– – 100.00%

0.0016 0.0913 − 0.0327 2.5781

0.0246 0.0746 − 0.5899 3.3665

– 100.00% − 43.66%

– – 100.00%

considerably from large to small-size stock portfolios for the markets of South Africa and sub-Saharan Africa. This is also reflected in the measure of volatility, where standard deviations increase dramatically from larger size firm to smaller size firm portfolios. Average returns in small-size stock portfolios tend to be more risky than in larger stock portfolios, but also have higher potential returns. However, the negative value of the mean of the SMB in Table 6 for North Africa and sub-Saharan Africa indicates the likelihood of a reverse size effect found by Fama and French (1993) where returns steadily decrease as stock size increases. Although there is little difference between the low and high liquidity portfolio means across the various market variables, there is an increase in volatility from high illiquidity to low illiquidity stock portfolios. This result is expected given that the often severe illiquidity inhibits price adjustment and returns in reaction to the impact of sudden erratic order flow on stock prices. The evidence in Table 6 shows that there is little correlation between the SMB, ILLIQ and market valuation factors for the market variables in South Africa and sub-Saharan Africa. In contrast, there is some correlation between the illiquidity and market factors in the London, Paris and North African markets. It should be noted that these differences indicate that the implicit assumption of integration on either an intra or inter market basis is tenuous at best. However, the lack of viable alternative methods and the ease of application merit the continued use of this approach. Table 7 demonstrates the high degree of non-normality for the aggregate market, top tier stocks and industrial sectors. Excess kurtosis is particularly high for Nigeria (45.2449), Zambia (41.4781) and Botswana (75.9508) while there is considerable contrast in the standard deviation, with values of over 15% for Tunisian top tier stocks, Nigerian Consumer non-cyclical, Energy and Overall segments, the Botswana overall market, and the financial sector in Zambia. Interestingly, South Africa's top tier stocks have an

270

B. Hearn, J. Piesse / Emerging Markets Review 10 (2009) 257–278

Table 7 Summary statistics for market and sector portfolios for period 2002 to 2008. Market

Industrial sector

Mean

Std. dev.

Skewness

Ex. kurtosis

Egypt

Financials Basic materials Consumer non-cyclical Industrial Overall Top 10 stocks Financials Consumer-cyclical Diversified Industrial Overall Top 10 stocks Financials Consumer-cyclical Consumer non-cyclical Overall Top 5 stocks Financials Consumer non-cyclical Energy Overall Top 10 stocks Financials Consumer non-cyclical Industrial Overall Top 5 stocks Financials Consumer non-cyclical Overall Top 3 stocks Financials Consumer non-cyclical Industrial Overall Top 3 stocks Financials Overall Financials Communications Basic materials Consumer-cyclical Consumer non-cyclical Energy Industrial Overall Top 10 stocks

2.65% 2.49% 2.09% 1.64% 2.08% 2.55% 2.09% 1.93% 2.01% 0.49% 1.65% 1.73% 0.92% 0.41% 2.71% 0.85% 3.02% 2.48% 3.68% 4.87% 4.51% 2.76% 3.50% 1.42% 2.25% 2.46% 3.55% 1.19% 0.84% 3.01% 1.71% 3.37% 5.00% 1.29% 4.29% 3.38% 1.23% 1.18% 1.66% 2.15% 2.25% 2.29% 2.35% 1.82% 2.30% 2.04% 4.11%

8.24% 9.55% 7.65% 8.62% 6.93% 7.86% 5.72% 7.75% 6.14% 3.66% 4.05% 5.47% 3.70% 5.90% 11.68% 3.29% 17.79% 8.56% 16.44% 17.14% 15.01% 9.42% 9.46% 6.60% 7.17% 6.59% 9.33% 3.68% 6.16% 17.68% 4.79% 17.44% 9.31% 6.26% 14.86% 10.61% 7.25% 5.49% 6.86% 7.48% 8.47% 8.21% 7.67% 7.92% 8.03% 6.87% 32.84%

0.337 0.854 0.332 1.649 0.260 0.170 1.240 1.603 0.538 − 0.569 0.393 0.259 0.534 1.356 1.646 0.236 2.212 1.040 5.918 4.102 5.923 2.015 0.496 0.469 1.218 0.008 0.532 0.062 − 0.041 8.482 0.282 5.564 0.893 0.786 5.526 0.216 1.227 1.002 − 0.601 − 0.206 − 0.816 − 0.119 −0.221 − 0.306 − 0.684 − 0.697 −1.402

3.488 3.839 3.851 8.314 3.376 2.885 5.279 10.658 4.146 3.804 3.348 4.601 3.967 10.096 8.273 2.857 14.783 4.725 46.730 26.437 45.245 14.392 4.375 4.222 8.739 4.317 4.128 2.887 5.335 75.95 3.407 41.171 5.480 4.973 41.478 3.603 8.151 5.964 3.762 3.850 5.890 3.856 3.359 4.901 4.147 4.013 12.962

Morocco

Tunisia

Nigeria

Kenya

Botswana

Zambia

Namibia South Africa

exceptionally high volatility (32.84%), which probably reflects the depth of markets and their ability to properly reflect investor sentiment and uncertainty over the volatile macroeconomic climate experienced during the sample period. However, the top tier stocks generally show considerable decreases in levels of skewness and kurtosis with the distribution of returns closer to normality than for the overall market. 4.2. Performance of traditional CAPM against three-factor CAPM Table 8 reports the results from the pooled regression on all nine size–illiquidity sorted portfolios for each of the market variables: North, South and sub-Saharan Africa (SSA). For the South African market

B. Hearn, J. Piesse / Emerging Markets Review 10 (2009) 257–278

271

Table 8 Time series regressions using equally weighted monthly contemporaneous market excess returns for 9 portfolios formed on size and illiquidity for period: 2002−2008, for all sample markets. Portfolio

S/L

S/M

S/H

Market: North Africa Panel A: CAPM-adjusted performance α̂(%) −0.0053 0.0073 − 0.0067 (−1.12) (1.07) (−1.39) β̂ 1.5189 0.5866 0.4746 (12.52) (4.00) (5.80) 0.6122 0.1826 0.2608 Adj R2(1) Panel B: three-factor CAPM performance α̂ 0.0011 0.0098 −0.0045 (0.31) (1.81) (− 1.58) β̂ 0.8786 0.9739 0.8761 (7.02) (6.69) (8.33) ŝ 0.2158 0.1744 0.1560 (4.85) (2.82) (6.68) ĥ −0.1969 0.1200 0.1243 (−4.66) (3.36) (5.23) 0.8004 0.3717 0.6421 Adj R2(4) Market: sub-Saharan Africa (ex. RSA) Panel A: CAPM-adjusted performance α̂(%) −0.0007 0.0137 0.0062 (− 0.07) (1.33) (0.78) β̂ 1.2760 1.1226 0.0493 (3.79) (2.19) (0.42) 0.3663 0.1986 0.0021 Adj R2(1) Panel B: three-factor CAPM performance α̂ −0.0041 0.0025 0.0037 (−0.49) (0.27) (0.5811) ̂β 1.3331 1.6047 0.1859 (5.76) (3.88) (1.21) ŝ 0.0136 0.3275 0.1019 (0.28) (3.05) (2.37) ĥ −0.1277 0.1757 0.1034 (−3.16) (2.19) (3.09) 0.5992 0.5749 0.1999 Adj R2(4) Market: South Africa Panel A: CAPM-adjusted performance α̂(%) 0.0036 0.0128 0.0160 (1.33) (3.26) (2.84) β̂ 1.1600 1.0111 0.7519 (23.73) (18.66) (8.53) 2 0.9050 0.8146 0.4713 Adj R (1) Panel B: three-factor CAPM performance α̂ −0.0005 0.0054 0.0036 (− 0.19) (1.8559) (1.03) β̂ 1.0381 0.9515 1.0245 (26.45) (26.097) (13.19) ŝ 0.1659 0.2006 0.1065 (5.71) (4.81) (2.71) ĥ − 0.2365 −0.1216 0.5050 (−5.71) (− 2.08) (8.39) 0.9437 0.8707 0.8533 Adj R2(4)

M/L

M/M

M/H

B/L

B/M

B/H

−0.0114 (−2.45) 2.0367 (12.74) 0.7529

0.0031 (0.95) 0.9615 (10.73) 0.5117

− 0.0114 (− 2.45) 2.0367 (12.74) 0.7529

0.0025 (0.44) 1.0755 (7.92) 0.5823

0.0082 (2.57) 1.0531 (12.93) 0.6113

0.0021 (0.57) 0.4406 (4.17) 0.2238

− 0.0092 (− 2.54) 1.2376 (9.58) − 0.0114 (− 0.35) −0.2466 (−6.82) 0.8453

0.0021 (0.58) 1.1247 (7.30) −0.0263 (− 0.54) 0.0503 (1.61) 0.5155

0.0020 (0.68) 1.0257 (9.31) 0.0035 (0.14) 0.1692 (7.30) 0.4822

− 0.0001 (−0.03) 0.7401 (6.66) − 0.1707 (− 6.79) −0.1039 (− 3.02) 0.7408

0.0049 (1.65) 1.0269 (9.02) − 0.1563 (− 7.36) − 0.0085 (− 0.30) 0.7140

− 0.0021 (−0.79) 0.9581 (9.39) −0.1262 (−7.21) 0.1593 (6.40) 0.6395

0.0046 (0.54) 0.7483 (2.67) 0.1788

0.0003 (0.04) 1.6692 (3.39) 0.2764

0.0168 (2.38) 0.2420 (1.97) 0.0515

0.0027 (0.30) 0.7018 (3.44) 0.2607

0.0018 (0.20) 0.6507 (5.29) 0.1953

− 0.0346 (− 0.96) 2.71 (1.51) 0.2044

0.0054 (0.75) 0.6329 (2.74) − 0.0999 (−2.97) −0.1686 (− 6.77) 0.4135

− 0.0032 (−0.42) 1.7579 (3.17) 0.0423 (0.39) − 0.0736 (− 0.98) 0.32

0.0157 (2.18) 0.3034 (1.94) 0.0461 (1.23) 0.0483 (1.47) 0.0904

0.0048 (0.76) 0.5402 (3.57) − 0.1288 (− 4.18) −0.1710 (− 6.75) 0.6313

0.0035 (0.41) 0.5614 (4.70) − 0.0652 (− 1.67) − 0.0589 (− 2.14) 0.2224

− 0.0097 (− 0.95) 2.0195 (12.90) − 0.3630 (− 8.09) 0.3812 (10.85) 0.9409

− 0.0054 (−2.08) 1.0633 (28.84) 0.9283

− 0.0042 (− 1.69) 1.0429 (38.86) 0.9189

− 0.0015 (−0.76) 0.9571 (29.30) 0.9073

− 0.0062 (− 1.63) 0.9873 (21.60) 0.8376

−0.0069 (− 2.12) 0.9979 (27.59) 0.8996

− 0.0041 (− 1.14) 0.9659 (18.95) 0.9111

− 0.0035 (−1.51) 0.9976 (24.48) − 0.0026 (−0.10) − 0.1228 (−3.86) 0.9427

−0.0043 (− 1.57) 1.0146 (34.59) 0.0179 (0.59) − 0.0539 (− 1.46) 0.9190

− 0.0035 (−1.51) 0.9665 (22.23) 0.0369 (1.36) 0.0157 (0.41) 0.9103

0.0026 (0.81) 0.9697 (14.05) − 0.1832 (− 5.41) −0.0239 (− 0.53) 0.9215

0.0002 (0.16) 1.0333 (46.59) −0.1802 (− 7.64) 0.0751 (3.23) 0.9576

0.0021 (0.89) 1.0128 (28.75) − 0.1635 (− 7.61) 0.0959 (3.36) 0.9586

Notes: (1) Numbers in parentheses are t-statistics. (2) One month T-bill risk free rate for month t, which is taken as the one month UK-Gilt rate in this case.

variables there is little increase in explanatory power between the one-factor CAPM and the three-factor alternative. This indicates both a considerable amount of intra-market integration and that the cross section of returns is sufficiently explained by a single market premium as opposed to additional size and

272

B. Hearn, J. Piesse / Emerging Markets Review 10 (2009) 257–278

liquidity factors. The additional factors in all three markets also act to further reduce the statistical significance of the Jensen alpha, αp, in the regressions of small-size high illiquidity, medium size low illiquidity and large-size medium illiquidity portfolios. However, the results for the North Africa and SSA market variables are markedly different from those in South Africa. In North Africa there are significant increases in explanatory power when the size and liquidity factors are included and generally the Jensen alpha, αp, terms are not statistically different from zero, indicating a good fit with established theoretical CAPM assumptions. However, in the case of SSA the explanatory power is frequently less than those of North Africa. The severe illiquidity affecting the model is highlighted in the adjusted R2 of 0.2% for the small-size high illiquidity portfolio in the one-factor model, which jumps 19.99% when the size and liquidity factors are included. A similar dramatic increase in explanatory power arises in the large-size high illiquidity portfolio where the adjusted R2 in the one-factor model is 20.44% and increases to 94.09% in the augmented version. Although the application of this model to highly illiquid markets is questionable, and the implicit assumptions regarding inter and intra asset market integration are tenuous, these are important results in the context of emerging markets, as the vast majority of research on the original CAPM (Sharpe, 1964; Lintner, 1965) has been restricted to developed markets only. In all cases within Africa the Jensen alpha, αp, term was not statistically different from zero which is in line with theory. The estimated coefficients on both the market excess return (β̂) and the illiquidity factor (HML) are large and significant in almost all cases. Those on the size factor-mimicking portfolio (SMB) are smaller in the majority of cases and are only significantly different from zero in the large or small-size company portfolios. The coefficients on the large-size portfolios are negative and statistically significant. The negative sign on the large-size portfolio betas indicates that large firms' returns decrease when the size premium increases, which is the opposite for small firms. This behaviour is not expected and is a contradiction of the well documented “size effect” in the valuation of smaller firms (Martinez et al., 2005) and is also a function of the highly heterogeneous universe of stocks in emerging markets, noted in Table 3. This conflicts with the expected result and does not provide investors with good hedging opportunities. Thus, as with the results for the small-size portfolios, a different valuation method would be needed to price highly illiquid stocks accurately. The estimated coefficients on the illiquidity factor-mimicking portfolios are negative for low and medium illiquidity portfolios indicating that more liquid firms experience a decrease in expected returns when aggregate market illiquidity increases. In general, the coefficients on the low illiquidity and medium illiquidity portfolios are negative, as one would expect, with firms paying lower returns when the illiquidity variable increases. However, the coefficients on the high illiquidity portfolios are positive indicating that these companies pay higher returns when the illiquidity measure increases. The increased explanatory power of these models illustrates that the augmented CAPM is appropriate for illiquid markets and thus appropriate in this context (Table 9). 4.3. Modelling country and industry portfolios and cost of equity estimation Table 10 reports estimates of the cost of equity calculated from the expected returns for each country and industry regression. It should be noted that the estimates for each market are given alongside the market universe from which they have been calculated. The high cost of equity for these markets is used as the discount factor and applied to future cash flows in project valuation. The cost of equity is calculated from the annualised combination of the total risk premium, which is the sum of market, size and illiquidity premiums, with the 3 month UK Treasury rate a proxy for the risk free rate. 4.3.1. Average returns in North Africa The inclusion of the size and liquidity factors within the three-factor model causes increases in explanatory power in all cases. Interestingly, in almost all industries in all three North African countries the coefficients on the size premium are small and frequently statistically insignificant, whereas those on the liquidity premium are large and significant. This supports the claim that liquidity is an important variable driving the returns process. However, the effect of severe illiquidity and price rigidity reduces the explanatory power of the model, such as in the Tunisian consumer non-cyclical industry. The adjusted R2 of the one-factor CAPM is less than 1% and rises to a meagre 4.24% when the size and liquidity factors are

B. Hearn, J. Piesse / Emerging Markets Review 10 (2009) 257–278

273

Table 9 Pooled cross-section regression for equally weighted monthly excess returns on country portfolios with size and illiquidity for 1996 to 2007. Finance

Comm.

Basic Cons. materials cyclical

Market: North Africa Egypt Panel A: CAPM-adjusted performance α̂(%) −0.0016 – − 0.0035 – (−0.33) (− 0.43) β̂ 1.6877 – 1.7002 – (13.81) (7.64) 2 – 0.4734 – Adj R (1) 0.6299 Panel B: three-factor CAPM performance α̂ 0.0003 – − 0.0052 – (0.05) (− 0.61) β̂ 0.9806 – 1.2987 – (4.98) (4.78) ŝ − 0.0053 – − 0.1364 – (−0.07) (− 1.77) ĥ −0.2183 – − 0.1243 – (− 4.58) (− 2.17) – 0.5079 – Adj R2(4) 0.7150 Tunisia Panel A: CAPM-adjusted performance α̂(%) 0.0034 – – 0.0020 (0.79) (0.31) β̂ 0.3467 – – 0.1223 (3.61) (0.83) – – 0.0065 Adj R2(1) 0.1221 Panel B: three-factor CAPM performance α̂ 0.0031 – – 0.0022 (0.79) (0.35) ̂β 0.6996 – – 0.4482 (5.09) (2.15) ŝ 0.0385 – – 0.0524 (1.23) (0.97) ĥ 0.1090 – – 0.1007 (3.33) (2.51) – – 0.0178 Adj R2(4) 0.2292 Morocco Panel A: CAPM-adjusted performance α̂(%) 0.0079 – – 0.0054 (1.53) (0.89) β̂ 0.7742 – – 0.8284 (6.47) (4.97) – – 0.1628 Adj R2(1) 0.2679 Panel B: three-factor CAPM performance α̂ 0.0048 – – 0.0029 (1.05) (0.44) β̂ 1.1887 – – 1.3744 (7.01) (3.21) ŝ −0.0889 – – −0.0431 (− 2.06) (−0.66) ĥ 0.1277 – – 0.1684 (3.51) (1.94) 2 – – 0.2097 Adj R (4) 0.3478 Market: sub-Saharan Africa (excl. RSA) Nigeria Panel A: CAPM-adjusted performance α̂(%) −0.0110 – – (− 1.56) ̂β 1.0856 – – (4.35) 2 – – Adj R (1) 0.4385

– – –

Cons. noncyclical

Diversified Energy

− 0.0075 – (− 1.54) 1.6981 – (13.15) 0.7413 – − 0.0053 (− 1.09) 1.3305 (9.64) 0.0537 (1.07) − 0.1133 (− 4.70) 0.7708

–– –– ––

Ind.

Overall

Top stocks

− 0.0089 (− 1.39) 1.5123 (6.61) 0.4591

− 0.0064 (− 1.95) 1.6370 (22.44) 0.8451

0.0020 (0.24) 1.4139 (10.38) 0.4849

− 0.0046 (− 1.58) 1.2431 (16.34) 0.0268 (1.05) − 0.1215 (− 6.85) 0.8849

− 0.0015 (− 0.25) 0.9669 (4.18) -0.2291 (− 3.39) − 0.1385 (− 2.77) 0.6186

0.0039 (1.06) 0.2801 (2.97) 0.0995

0.0157 (0.72) 0.8783 (2.00) 0.0251 0.0121 (0.59) 1.6937 (2.28) − 0.0533 (− 0.35) 0.2515 (1.72) 0.0639

–

––

–

––

–

––

–

––

–

––

− 0.0079 (− 1.24) 1.2502 (4.42) 0.0069 (0.14) − 0.0809 (− 1.38) 0.4575

0.0274 – (1.82) − 0.0168 – (− 0.05) 0.00003 –

––

––

––

––

––

––

0.0301 (1.63) 0.6181 (1.57) 0.2165 (1.65) 0.1966 (2.18) 0.0424

–

––

––

–

––

––

–

––

––

–

––

––

–

––

––

0.0040 (1.08) 0.6095 (5.25) 0.0517 (1.74) 0.1018 (4.07) 0.2358

–

0.0119 (1.88) 0.4887 (3.31) 0.0841

––

-0.0022 (− 0.54) 0.4279 (3.99) 0.1972

0.0078 (1.84) 0.5290 (6.07) 0.2501

0.0083 (1.32) 0.5497 (3.78) 0.1432

− 0.0050 (-1.19) 0.7276 (5.84) -0.0911 (− 3.43) 0.0922 (2.73) 0.3422

0.0046 (1.31) 0.9247 (7.57) − 0.0961 (− 2.99) 0.1219 (4.40) 0.4267

0.0029 (0.58) 0.8755 (5.69) − 0.2029 (− 4.54) 0.0999 (2.86) 0.3423

––

-0.0101 (− 1.49) 1.2369 (4.11) 0.5052

0.0018 (0.18) 0.7865 (4.44) 0.1836

– – – – – – –

0.0084 (1.54) 0.7329 (3.44) − 0.1348 (− 3.26) 0.0749 (1.71) 0.1434

− 0.0152 – (−1.28) 1.5752 – (2.45) 0.2451 –

–– –– –– –– –– –– ––

0.0110 (0.59) 1.1409 (3.23) 0.1118

–– ––

(continued on next page)

274

B. Hearn, J. Piesse / Emerging Markets Review 10 (2009) 257–278

Table 9 (continued) Finance

Comm.

Basic Cons. materials cyclical

Cons. noncyclical

Diversified Energy

Ind.

Overall

Top stocks

−0.0232 (− 1.42) 1.9059 (2.46) 0.2208 (1.07) 0.0971 (0.64) 0.3498

–

––

––

-0.0127 (− 1.96) 1.3054 (4.92) 0.0339 (0.56) − 0.0491 (− 1.19) 0.6029

0.0052 (0.57) 0.5949 (4.05) − 0.1430 (− 3.02) − 0.1456 (− 4.74) 0.3313

0.0141 (1.54) 0.2557 (1.12) 0.0235

0.0072 (1.097) 0.5306 (2.42) 0.1691

0.0151 (1.39) 0.6201 (2.12) 0.1112

0.0059 (1.09) 0.5384 (2.49) -0.0084 (− 0.21) − 0.0777 (− 2.88) 0.3369

0.0145 (1.55) 0.5723 (2.57) − 0.0536 (− 1.16) − 0.1421 (− 3.79) 0.3052

-0.0152 (− 0.73) 1.3768 (1.26) 0.1578

0.0141 (2.57) 0.0952 (0.68) 0.0109 0.0136 (2.21) 0.1325 (0.82) 0.0297 (0.80) 0.0393 (1.33) 0.0193

Market: sub-Saharan Africa (excl. RSA) Panel B: three-factor CAPM performance α̂

− 0.0129 – – – (−2.33) β̂ 1.1149 – – – (7.54) ŝ 0.0048 – – – (0.21) ĥ − 0.0784 – – – (− 4.11) – – – Adj R2(4) 0.5690 Kenya Panel A: CAPM-adjusted performance α̂(%) 0.0111 – – – (1.22) β̂ 0.7239 – – – (2.73) – – – Adj R2(1) 0.1519 Panel B: three-factor CAPM performance α̂ 0.0102 – – – (1.37) β̂ 0.6862 – – – (3.22) ŝ −0.0470 – – – (− 0.87) ĥ −0.1398 – – – (−3.61) 2 – – – Adj R (4) Botswana Panel A: CAPM-adjusted performance α̂(%) 0.0068 – – – (1.73) β̂ 0.1570 – – – (1.58) – – – Adj R2(1) 0.0388 Panel B: three-factor CAPM performance α̂ 0.0068 – – – (1.73) β̂ 0.1553 – – – (1.39) ŝ −0.0007 – – – (− 0.04) ĥ 0.0022 – – – (0.13) 2 – – – Adj R (4) 0.0157 Zambia Panel A: CAPM-adjusted performance α̂(%) − 0.0044 – – – (−0.37) β̂ 1.1559 – – – (1.74) 2 – – – Adj R (1) 0.1109 Panel B: three-factor CAPM performance α̂ −0.0062 – – – (−0.55) β̂ 1.2466 – – – (1.93) ŝ 0.0657 – – – (0.91) ĥ 0.0573 – – – (0.88) – – – Adj R2(4) 0.0974

– – – –

− 0.0008 – (− 0.12) 0.4543 – (2.38) 0.1207 – − 0.0029 (− 0.47) 0.5134 (2.37) 0.0307 (0.75) − 0.0339 (− 1.17) 0.2240

0.0139 (0.66) 0.9848 (2.80) − 0.1145 (− 1.44) − 0.1064 (− 1.86) 0.1188 –– –– ––

–– –– ––

–

––

–

––

–

––

–

––

–

––

0.0122 (1.60) 0.3132 (1.28) 0.0335 (0.55) -0.0115 (− 0.23) 0.0498

0.0084 – (1.01) 1.15E-05 – (0.01) 0.0000 –

––

––

––

––

––

––

0.0091 (1.08) −0.0257 (− 0.19) −0.0152 (− 0.42) 0.0037 (0.12) 0.0119

–

––

––

–

––

––

–

––

––

–

––

––

–

––

––

-0.0009 (-0.16) 0.9846 (5.039) -0.2039 (-6.42) 0.2262 (9.62) 0.9078

0.0376 (2.58) 0.3723 (1.68) 0.0326

–

––

–

––

–

––

0.0086 (0.95) 0.1301 (0.96) 0.0119

0.0099 (1.22) 1.0028 (1.82) 0.1157

0.0126 (0.87) 0.6445 (2.99) 0.0916

0.0349 (2.25) 0.5232 (1.96) 0.1111 (1.86) 0.1057 (2.14) 0.1037

–

––

–

––

–

––

–

––

–

––

0.0078 (0.83) 0.1929 (1.22) 0.0501 (1.56) 0.0661 (2.37) 0.0499

0.0052 (0.51) 1.2354 (1.81) 0.1653 (1.46) 0.1275 (1.58) 0.1656

0.0106 (0.69) 0.7758 (3.11) 0.1022 (1.49) 0.1244 (2.13) 0.1608

B. Hearn, J. Piesse / Emerging Markets Review 10 (2009) 257–278

275

Table 9 (continued) Finance

Comm.

Basic Cons. materials cyclical

Market: South Africa Namibia Panel A: CAPM-adjusted performance α̂(%) −0.0007 – – – (− 0.11) β̂ 0.5249 – – – (5.56) 2 – – – Adj R (1) 0.2836 Panel B: three-factor CAPM performance α̂ −0.0035 – – – (−0.58) β̂ 0.6070 – – – (4.91) ŝ 0.0117 – – – (0.19) ĥ 0.1529 – – – (1.42) – – – Adj R2(4) 0.3014 South Panel A: CAPM-adjusted performance Africa α̂(%) −0.0049 − 0.0013 0.0018 − 0.0028 (− 2.15) (− 0.38) (0.20) (− 0.68) β̂ 0.8744 0.9255 0.8384 1.0439 (24.78) (20.87) (6.59) (23.55) 0.8509 0.5395 0.8996 Adj R2(1) 0.9034 Panel B: three-factor CAPM performance α̂ − 0.0047 − 0.0015 0.0105 − 0.0038 (−1.83) (−0.40) (1.31) (− 0.9431) β̂ 0.8231 0.9157 0.9638 0.9923 (24.55) (18.46) (8.08) (17.65) ŝ 0.0254 0.0109 −0.2665 0.0533 (1.25) (0.29) (− 4.12) (1.88) ĥ −0.0974 − 0.0187 0.2482 − 0.0994 (−2.76) (− 0.39) (2.52) (−2.09) 2 0.8476 0.6281 0.9044 Adj R (4) 0.9109

Cons. noncyclical

Diversified Energy

Ind.

Overall

Top stocks

–

–

––

––

––

–

–

––

––

–

–

––

––

0.0015 (0.33) 0.4229 (6.58) 0.3221

–

–

––

––

–

–

––

––

–

–

––

––

–

–

––

––

–

–

––

––

0.0002 (0.05) 0.4802 (5.52) -0.0065 (− 0.14) 0.1075 (1.34) 0.3295

− 0.0008 – (− 0.29) 0.9858 – (27.87) 0.9181 –

0.0042 (0.52) 0.5683 (4.55) 0.2783

-0.0022 (-0.87) 1.0229 (31.95) 0.9017

-0.0021 (-2.92) 0.9179 (39.78) 0.9912

-0.0098 (-0.22) 2.0719 (2.67) 0.2121

−0.0006 (−0.21) 0.9834 (20.77) −0.0020 (− 0.07) −0.0043 (− 0.08) 0.9161

0.0133 (2.02) 0.6911 (5.69) -0.2749 (-4.14) 0.2436 (2.02) 0.3819

-0.0064 (-2.34) 1.0712 (28.19) 0.0620 (2.59) 0.0872 (2.30) 0.9267

-0.0015 (-2.09) 0.9252 (37.69) -0.0175 (-1.68) 0.0143 (1.14) 0.9916

0.0276 (0.83) 2.0772 (2.43) -0.8277 (-1.45) 0.0509 (0.10) 0.2848

– – – – –

–– –– –– –– –– –– ––

Notes: (1) The risk free rate is the three month UK treasury/Gilt rate adjusted for monthly values. (2) Numbers in parentheses are Newey–West HAC covariance adjusted t-statistics.

included. Although this represents a sizeable jump in absolute terms it does question the use of OLS estimation in modelling highly illiquid series. In general, the adjusted R2 indicates the model has reasonable explanatory power across the North African markets. However, despite the apparent fit of the model there are some inconsistencies in the estimates of the cost of equity, in Table 10. As expected, Egypt as the least developed market, has the highest cost of equity across all industry sectors and even the top tier stocks. Costs of equity are greater than 30% for all sectors and 33.52% for the top tier blue chip stocks. The more developed Moroccan market has a wider dispersion of cost of equity between the various industrial sectors. This ranges from 15.43% and 17.26% for the industrial and diversified sectors to 25.37% for the consumer-cyclical sector. The top tier stocks have a cost of equity of 20.88% compared to an aggregate market value of 18.63%. Finally, the highly developed Tunisian market exhibits the greatest dispersion of cost of equity between sectors. Values range from 1.70% for the consumer non-cyclical industry to 11.19% for the finance sector. However, the greatest difference is between the top stocks (29.13%) and the overall market (9.21%). These results support Ben Naceur and Chaibi (2005) although the wide dispersion of cost of equity estimates in this case does question both the applicability of this pricing model to the very small and highly illiquid Tunisian market and the controversial underlying assumption of an integrated North African market universe. This is especially an issue given that a significant proportion of trading in Tunisia is undertaken by call auction which is fundamentally different from the continuous systems of Morocco and Egypt.

276

B. Hearn, J. Piesse / Emerging Markets Review 10 (2009) 257–278

Table 10 Cost of equity estimates derived from multifactor regression (%). Market sector

Finance Communications Basic materials Consumer-cyclical Consumer non-cyclical Diversified Energy Industrial Top stocks Overall

North Africa

Sub-Saharan Africa (Excl. RSA)

South Africa

Egypt

Tunisia

Morocco

Nigeria

Kenya

Zambia

32.13

11.19

24.03

55.41

36.33

59.22

Botswana 8.62

Namibia

South Africa

22.95

0.11

30.77 33.03 15.60 38.47 34.44

5.69 1.70

25.37 88.71

25.92

23.69

16.03 30.99 28.07

9.06 35.53 56.83

6.74 46.69

5.75 42.11 13.36 31.41

38.13 32.08

17.26 50.66 32.00 33.52 33.11

29.13 9.21

15.43 20.88 18.63

32.90 63.82

17.28

Notes: (1) Annualised cost of equity estimates generated at 12/2008 from the total risk premium. (2) The UK 3 month Gilt/Treasury rate is used in each case for risk free rate. (3) Top stocks refers to the top stocks as ranked by market capitalization from the overall market universe. It refers to the top 10 stocks for Egypt, Morocco and South Africa, the top 5 stocks for Kenya, Botswana and Tunisia, and the top 3 stocks for Zambia. The numbers of top stocks in each case are chosen on criteria of data availability and the number of stocks in the overall universe.

4.3.2. Average returns in the sub-Saharan African markets As with the North African markets, the inclusion of the additional size and liquidity factors causes increases in explanatory power in both the overall and top tier stocks in all sub-Saharan markets. The greatest increases arise from the added size factor in the overall markets of Botswana, which causes a jump in adjusted R2 of 15.77% to 90.77% and for Nigerian top tier stocks where the adjusted R2 jump from 18.36% to 33.13% following the inclusion of the liquidity factor. In general, the greatest difference in the levels of explanatory power arises within Nigeria and Botswana where the adjusted R2 for the top tier stocks is considerably lower than those for the overall market. This suggests that the high degree of segmentation and the concentrated profile of these markets between the large and liquid blue chip stocks and the small and severely illiquid remainder. Compared with the evidence from the aggregate market the profiles of the top tier stocks are quite different. Top tier stock returns across the sample appear to be better explained by the addition of the liquidity factor alone with the size factor statistically insignificant. However, despite the segmentation within the sub-Saharan market universe the levels and differences in the cost of equity are as expected. The aggregate Nigerian market has the highest at 63.82% that drops to 32.90% for the top tier stocks. Botswana and Zambia, which have similar size differences in intra-market illiquidity also have lower costs of equity in the overall market for the top stocks. Botswana falls from an aggregate value of 46.69% to 6.74%, and Zambia from 56.83% to 35.53%. The very low levels of top stocks in Botswana is probably due to the presence of many blue chip South African mining and finance companies that have secondary listings on the Gabarone exchange. These adhere to the high regulatory and governance standards in place in their primary listing on the Johannesburg Stock Exchange. In addition, the financial sectors in Botswana, Zambia, and Kenya have the highest costs of equity of all industries with the sole exception of Nigeria where the consumer non-cyclical industry (88.71%) surpasses the financial sector (55.41%). 4.3.3. Average returns in South Africa and Namibia Table 9 reports the results from South Africa and Namibia and again provides evidence of intra-market segmentation. There is a general decrease in the adjusted R2 between the overall South African market and South African top tier stocks as well as with the overall Namibian market. Generally the explanatory power of the one-factor CAPM is sufficient with only incremental increases in adjusted R2 after including the size factor in South African top stocks and the liquidity factor in Namibia. The cost of equity in the aggregate South African market is 31.41% which significantly decreases to 13.36% for the top ten stocks. In contrast, Namibia has a value of 17.28%. While these values are high, Correia and Uliana (2004) find costs of equity using a one-factor CAPM on a similar scale. However, there is considerable dispersion in the cost of equity

B. Hearn, J. Piesse / Emerging Markets Review 10 (2009) 257–278

277

between sectors, with the energy (5.75%) and basic materials (15.60%) sectors in South Africa having the lowest values and consumer-cyclical (38.47%) and industrial (42.11%) having the highest. 5. Conclusions This study proposes a size and liquidity augmented capital asset pricing model to explain the cross section of expected returns in emerging markets and is the first such study on a sample of African markets. The sample includes Morocco, Tunisia and Egypt in North Africa, Kenya, Botswana, Nigeria and Zambia in sub-Saharan Africa plus Namibia and South Africa. There are considerable differences between markets in terms of corporate governance and regulation with the most developed regimes being those in the North Africa group and in South Africa. Illiquidity series were constructed on a time series cross-section basis and augment the Fama and French (1993) risk-adjusted CAPM. This model is then used to calculate cost of equity estimates on samples that include top tier stocks, industrial sectors and the aggregate markets for these countries. This study presents substantial evidence regarding the importance of firm size and liquidity in pricing state variables. Considerable improvements are made in explaining the cross section of stock returns by including the size and liquidity factors, which is of particular importance in emerging markets where illiquidity is a common issue. The evidence suggests that costs of equity for the aggregate markets are highest in Nigeria and Zambia but lower in South Africa and Egypt and lower still in Kenya and Botswana. The lowest costs of equity are in markets in Namibia, Tunisia and Morocco. The greatest difference in costs of equity between the top tier blue chip stocks and the aggregate markets in Botswana and South Africa although there are smaller differentials in Nigeria and Zambia. This illustrates the significant segmentation present within markets between the top level blue chip stocks, which can access international capital markets and commonly adhere to the highest levels of corporate governance, and the aggregate market. Equally, the frequently high costs of equity of the various national financial sectors in the sample demonstrate that this industry is not necessarily the best from which exchanges can actively seek to attract new listings. This evidence also lends further support to the benefits of effectively enforced regulation and corporate governance regimes where the markets of Namibia, South Africa, Tunisia and Morocco have the lowest costs of equity and the highest levels of regulation and market transparency. It is clear that firms raising external finance through the stock market in these countries are at a distinct advantage compared to those in Nigeria and Zambia where the very high costs of equity inhibit industrial expansion and make otherwise profitable projects unviable. Acknowledgement The authors would like to thank Kate Phylaktis for useful comments and advice. References Amihud, Y., 2002. Illiquidity and stock returns: cross section and time series effects. Journal of Financial Markets 5, 31–56. Barako, D.G., Hancock, P., Izan, H.Y., 2006. Relationship between corporate governance attributes and voluntary disclosures in annual reports: the Kenyan experience, financial reporting. Regulation and Governance 5 (1), 1–30. Ben Naceur, S., Chaibi, H., 2005. Asset pricing and cost of equity in the Tunisian banking sector: panel data evidence. Economic Notes 39 (1), 89–113. Bourse de Casablanca website, 2009, Bourse de Casablanca, Morocco. http://www.casablanca-bourse.com/ Accessed on 10 February 2009. Bourse de Tunis website, 2009, Bourse de Casablanca, Morocco. http://www.bvmt.com.tn/ Accessed on 10 February 2009. CASE website, 2009, http://www.egyptse.com/ Accessed on 10 February 2009. Collins, D., Abrahamson, M., 2006. Measuring the cost of equity in African financial markets. Emerging Markets Review 7, 67–81. Correia, C., Uliana, E., 2004. Market segmentation and the cost of equity of companies listed on the Johannesburg Stock Exchange. South African Journal of Accounting Research 18 (1), 65–86. Daniel, K., Titman, S., 1997. Evidence on the characteristics of cross sectional variation in stock returns. Journal of Finance 52, 1–33. Fama, E., French, K., 1993. Common risk factors in the returns on stocks and bonds. Journal of Financial Economics 33, 3–56. Fawzy, S, 2003, Assessment of Corporate Governance in Egypt, The Egyptian Centre for Economic Studies Working Paper 82. Hearn, B., Piesse, J., 2002. Equity market integration versus segmentation in three dominant markets of the southern African customs union: cointegration and causality tests. Applied Economics 34 (14), 1711–1722. Hearn, B., J. Piesse, 2009, Modelling size and illiquidity in West African equity markets. Unpublished Working Paper: February 2009, Department of Management, King's College London.

278

B. Hearn, J. Piesse / Emerging Markets Review 10 (2009) 257–278

Hearn, B., J. Piesse, R. Strange, 2009. Market liquidity and stock size premia in emerging financial markets: the implications for foreign investment. International Business Review 16. Jefferis, K., 1995. The Botswana share market and its role in financial and economic development. World Development 23, 663–678. JSE website, 2009, http://www.jse.co.za/ Accessed on 10 February 2009. Jun, S.-G., Marathe, A., Shawky, H.A., 2005. Liquidity and stock returns in emerging equity markets. Emerging Markets Review 4, 1–24. La Porta, R., Lopez-de-Silanes, F., Shleifer, A., 2008. The economic consequences of legal origins. Journal of Economic Literature 46, 285–332. Lavelle, K., 2001. Architecture of equity markets: the Abidjan Regional Bourse. International Organization 55, 717–742. Lesmond, D.A., 2005. Liquidity of emerging markets. Journal of Financial Economics 77, 411–452. Lintner, J., 1965. The valuation of risk assets and the selection of risky investments in stock portfolios and capital budgets. Review of Economics and Statistics 17, 13–37. Liu, W., 2006. A liquidity-augmented capital asset pricing model. Journal of Financial Economics 82, 631–671. Martinez, M.A., Nieto, B., Rubio, G., Tapia, M., 2005. Asset pricing and systematic liquidity risk: An empirical investigation of the Spanish stock market. International Review of Economics and Finance 14, 81–103. Mishra, D.R., O'Brien, T.J., 2005. Risk and ex ante cost of equity estimates of emerging market firms. Emerging Markets Review 6, 107–120. Nigerian Stock Exchange website, 2009, Nigerian Stock Exchange History, Available at http://www.nigerianstockexchange.com/ the_exchange.htm Accessed 14 July 2009. O'Hara, M., 2003. Presidential address: liquidity and price discovery. Journal of Finance 58, 1335–1354. Old Mutual, 2009, http://www.oldmutual.com/ Accessed on 21 March 2009. Pastor, L., Stambaugh, R., 2003. Liquidity risk and expected stock returns. Journal of Political Economy 111, 642–685. Piesse, J., Hearn, B., 2005. Regional integration of equity markets in Sub-Saharan Africa. South African Journal of Economics 73 (1), 36–52. Sharpe, W.F., 1964. Capital asset prices: a theory of market equilibrium under conditions of risk. Journal of Finance 19, 425–442. Shum, W.C., Tang, G.Y.N., 2005. Common risk factors in returns in Asian emerging stock markets. International Business Review 14, 695–717. South African Institute of Directors, 2009, King II corporate governance report, http://www.iodsa.co.za/king.asp Accessed 10 February 2009. Zribi, H., 2008, Interview with Hatem Zribi, Market Development, Bourse de Tunis, Tunisia. 18 January 2008.